Diagnostic and Prognostic Role of miR-192 in Different Cancers: A Systematic Review and Meta-Analysis

Introduction It has been shown that miR-192 is abnormally expressed in a variety of cancer types and participates in different kinds of signaling pathways. The role of miR-192 in the diagnosis and prognosis of cancer has not been verified. This article is aimed at exploring the diagnostic and prognostic value of miR-192 through a systematic review and meta-analysis. Methods A systematic search was performed through PubMed, Embase, Web of Science, and Cochrane Library databases up to June 16, 2020. A total of 16 studies were enrolled in the meta-analyses, of which 11 articles were used for diagnostic meta-analysis and 5 articles were used for prognostic meta-analysis. The values of sensitivity and specificity using miR-192 expression as a diagnostic tool were pooled in the diagnostic meta-analysis. The hazard ratios (HRs) of overall survival (OS) with 95 confidence intervals (CIs) were extracted from the studies, and pooled HRs were evaluated in the prognostic meta-analysis. Eleven studies including 667 cancer patients and 514 controls met the eligibility criteria for the diagnostic meta-analysis. Five studies including 166 patients with high miR-192 expression and 236 patients with low miR-192 expression met the eligibility criteria for the prognostic meta-analysis. Results The overall diagnostic accuracy was as follows: sensitivity 0.79 (95%CI = 0.75-0.82), specificity 0.74 (95%CI = 0.64-0.82), positive likelihood ratio 3.03 (95%CI = 2.11-4.34), negative likelihood ratio 0.29 (95%CI = 0.23-0.37), diagnostic odds ratio 10.50 (95%CI = 5.89-18.73), and area under the curve ratio (AUC) 0.82 (95%CI = 0.78-0.85). The overall prognostic analysis showed that high expression of miR-192 in patients was associated with positive survival (HR = 0.62, 95%CI : 0.41-0.93, p = 0.020). Conclusion Our results revealed that miR-192 was a potential biomarker with good sensitivity and specificity in cancers. Moreover, highly expressed miR-192 predicted a good prognosis for patients.


Introduction
Cancer is threatening human health and shorting human life. Around 1.8 million new cancer cases and 60 thousands of cancer deaths occurred in the United States based on the cancer statistics 2020 [1]. One reason for tumor death is that the tumor is already in its advanced stage as soon as it is discovered. In this case, it is important to find a marker that can detect tumors sensitively. However, a sensitive tumor biomarker is lacking in clinical practice. At present, more and more attention has been focused on microRNAs, which are highly conserved, short noncoding RNAs. MicroRNA binds to the 3 ′ untranslated region of target mRNA by base pairing matching, resulting in degradation of target mRNA or inhibi-tion of protein translation, which is involved in biological progress such as cell growth, differentiation, proliferation, and apoptosis [2]. Besides, miRNAs have been reported to regulate the key characteristics of cancer, involving selfsufficiency in growth signal, antigrowth signal, evasion from apoptosis, limitless replicative potential, angiogenesis, invasion, and metastasis [3]. Therefore, miRNAs could be promising biomarkers for diagnosis and prognosis [4].
MicroRNA-192 was firstly confirmed by Lim et al. in 2003 [5]. It is reported to be overexpressed in gastric cancer [6], hepatocellular carcinoma [7], and neuroblastoma [8], but downregulated in colorectal cancer [9] and lymphoblastic leukemia [10]. MicroRNA-192 plays a critical role in cell proliferation, migration and invasion [11], apoptosis [12], and epithelial-to-mesenchymal transition [13]. More importantly, miR-192 has been consistently detected in sputum [14], cervical cancer tissue, serum [15], and urine [16], suggesting that miR-192 might be a valuable biomarker for cancer diagnosis and detection. However, no meta-analyses concerning association between miR-192 expression and cancer diagnosis and prognosis have been published. Here, we conducted the diagnostic and prognostic meta-analyses to assess the diagnostic and prognostic value of miR-192.

Materials and Methods
2.1. Literature Search Strategy. A systematic literature search was carried out in PubMed, Embase, Web of Science, and Cochrane Library databases up to June 2020. The first part was to screen articles that explored the diagnostic value of miR-192 in cancers. Both MeSH terms and free-text words were used in the search strategy. The following search keywords were used in combination: "neoplasms" or "tumor" or "cancer", "diagnosis", "ROC curve", "sensitivity and specificity", and "microRNA-192". The second part was to screen articles that explored the prognostic value of miR-192 in cancers. The keywords were as follows: "neoplasms" or "tumor" or "cancer", "survival", "prognosis", "recurrence", and "microRNA-192".

Inclusion and Exclusion
Criteria. For diagnostic metaanalysis, studies were included for further evaluation if they meet the following criteria: (1) any types of cancers concerning miR-192, (2) inclusion of a diagnostic standard, (3) sufficient data (true positive, false positive, false negative, and true negative) for calculating the sensitivity and specificity, (4) studies based on humans, and (5) studies published in English. Exclusion criteria were as follows: (1) non-English articles; (2) other types of articles such as conference abstracts, reviews, meta-analysis, patents, case reports, comments, and letters; and (3) insufficient data for calculating the sensitivity and specificity.
For prognostic meta-analysis, studies were included for further evaluation if they meet the following criteria: (1) any types of cancers concerning miR-192, (2) inclusion of a diagnostic standard, (3) associations between the expression of miR-192 and prognosis of patients being determined, (4) hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) being evaluated, and (5) English publications. Exclusion criteria were as follows:   BioMed Research International   (1) non-English articles; (2) other types of articles such as conference abstracts, reviews, meta-analysis, patents, case reports, comments, and letters; (3) insufficient data to calculate the HRs and 95% CIs; and (4) the prognostic data based on TCGA dataset.

Data Extraction and Quality Assessment.
All studies were independently selected by two investigators (Lili Wang and Yuhan Liu), and uncertain data were reviewed by a third author (Chen Lyu). The following information was collected in diagnostic analysis: first author's name, publication year, nationality, ethnicity, cancer type, sample type, test method, cut-off, case number, sensitivity, and specificity. For the prognostic analysis, the following information was collected: first author's name, publication year, nationality, cancer type, cases of high expression of miR-192, cases of low expression of miR-192, the endpoint of follow-up, and HRs along with their corresponding 95% CIs. The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS 2) tool [17] was used to assess the quality of articles included in the diagnostic meta-analysis. The Newcastle-Ottawa Scale (NOS) [18] was used to assess the articles in the prognostic meta-analysis.

Statistical Analysis.
All data analyses were performed using Stata MP 16.0 software (StataCorp, College Station, TX). For the diagnostic meta-analysis, the pooled sensitivity, specificity, diagnostic odds ratio (DOR), positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were generated through bivariate meta-analysis. The summary receiver operator characteristic (SROC) curve and the area under the curve (AUC) were calculated to evaluate the overall diagnostic value of miR-192 in cancers. The heterogeneity test was conducted using the chi-square-based Q test and Higgins I-squared statistic. That I 2 > 50%, and p < 0:10 indicated heterogeneity among studies. Publication bias was evaluated by funnel plots and by Begg's and Deeks' tests. p < 0:05 suggests the existence of publication bias in studies. For the prognostic meta-analysis, the pooled HRs and 95% CIs were determined using the Z-test, with p < 0:05 defined as significant. HR > 1 indicated poor prognosis for patients with miR-192, while HR < 1 meant a protective effect for the prognosis of highly expressed miR-192. The methods for the assessment of heterogeneity and publication bias were the same as those for the diagnostic meta-analysis.

Quality Assessment of Studies.
The quality of diagnostic meta-analysis was assessed using the QUADAS-2 tool. All the studies were scored between 4 and 6 points which represented moderate or high quality (Table 1). For the prognostic meta-analysis, the Newcastle-Ottawa Scale (NOS) tool was used to assess the quality of studies according to three elements: selection (0-4 points), comparability (0-2 points), and outcome (0-3 points). All the studies were assessed as moderate or high quality, with scores between 5 and 7 points (Table 3).
3.3. The Results of the Diagnostic Meta-Analysis. The sensitivity and specificity of 11 studies are presented in the forest plots as shown in Figure 2. There was no heterogeneity in the sensitivity (I 2 = 9:63%, 95%CI = 0-61.53%), but significant heterogeneity in the specificity (I 2 = 76:64%, 95%CI = 63:03%-90.24%). Overall, the sensitivity and specificity for the pooled data were 0.79 (95%CI = 0:75-0.82) and 0.74 (95%CI = 0:64-0.82). In addition, the pooled PLR was 3.03 (95%CI = 2:11-4.34), and the NLR was 0.29 (95%CI = 0:23 -0.37) as shown in Figure 3. The DOR was 10.50 (95%CI = 5:89-18.73, Figure 4). The SROC curve is shown in Figure 5. The AUC for the miR-192 test method was 0.82 (95%CI = 0:78-0.85), suggesting that miR-192 has a relatively high diagnostic value. Fagan's nomogram was applied for assessing the clinical utility of the index test shown in Figure 6. When miR-192 was tested in patients with a pretest probability of cancer of 50%, the posttest probability of having cancer was improved to 75% by a positive result, while the posttest probability without cancer was dropped to 22% by a negative result. Taken together, miR-192 had a relatively moderate accuracy for identification of cancer patients.    Figure 4: Forest plot of the diagnostic odds ratio for miR-192 in the diagnostic meta-analysis.     (Figures 7(a) and 7(b)) analyses suggested that the bivariate model was moderately robust. Influence analysis (Figure 7(c)) and outlier detection (Figure 7(d)) did not identify any outliers.
3.6. Threshold Effect and Heterogeneity. The ROC plane showed the appearance of a nontypical shoulder arm suggesting no threshold effect existing (Figure 9). Spearman's correlation coefficient was -0.374 (p = 0:258), also indicating no threshold effect existing. The Galbraith radial plot showed that all the studies were in the 95% CI region suggesting no heterogeneity (Figure 10(a)). The bivariate boxplot showed that most studies were scattered in the middle region except three studies suggesting that there was heterogeneity between studies ( Figure 10(b)). Due to only 11 studies included in the diagnostic meta-analysis, it was difficult to perform the subgroup and meta-regression analyses to investigate the sources of heterogeneity.

A Prognostic Meta-Analysis of the Relationship between miR-192 Expression and Prognosis in Cancers.
Five studies were used to assess the OS shown in Figure 11. There was statistically significant heterogeneity (I 2 = 65:9%, p = 0:019), so a random effects model was used. The pooled HR was 0.62 (95%CI = 0:41-0.93, p = 0:020), suggesting that a high level of miR-192 was associated with positive patients' survival. The funnel plot was symmetrical, and Begg's test (p = 0:806) also indicated that there was no publication bias ( Figure 12).

Discussion
At present, imaging examination is used as a means of preliminary diagnosis and the final diagnosis still relies on the invasive biopsy. Once discovered, most cancers have entered the advanced stage, which causes great difficulties in treatment. Moreover, biopsy progress is invasive and may cause tumor dissemination [31]. Therefore, cancers might be detected at an early stage, if some biomarkers can be used for tumor screening clinically, providing the possibility of a cure. It is also possible to judge the prognosis by biomarkers, which is convenient, fast, and economical.
miRNA is a type of small noncoding RNA that plays important regulatory roles in gene expression and various biological processes [32]. Recently, miRNAs have been shown to have the potential to predict the diagnosis and prognosis of cancer patients [33], which could be used as diagnostic and prognostic biomarkers. miR-192 is one of those miRNAs reported to be a potential diagnostic and prognostic marker [23,28].
The pooled PLR was 3.03 (95%CI = 2:11-4.34), and NLR was 0.29 (95%CI = 0:23-0.37), meaning that cancer patients had a higher 3.03-fold probability of being miR-192 positive compared to control patients, and the probability of a negative result in patients was 0.29 times that in nonpatients. Fagan's nomogram revealed that when a pretest probability of 50% was specified, the positive posttest probability would increase to 75%, and the negative posttest probability would decrease to 22%. The results suggest that miR-192 is reliable for the detection and diagnosis in NSCLC, CCA, HCC, PDAC, PC, CC, BC, and AML.
In the diagnostic meta-analysis, the sensitivity had no heterogeneity (I 2 = 9:63%, 95%CI = 0-61.53%), but there  11 BioMed Research International was significant heterogeneity in the specificity (I 2 = 76:64%, 95%CI = 63:03%-90.24%). One reason for the heterogeneity perhaps was that the ethnicity of patients in most studies was Asian, which may result in the bias. Secondly, the cutoff values in studies were different and some of them were not mentioned. Thirdly, different sample types might contribute to the heterogeneity. However, the ROC plane represented a nontypical shoulder arm-like appearance, and Spearman's correlation coefficient was -0.374 (p = 0:258), suggesting that there was no threshold effect. Therefore, the threshold effect was not a major cause of heterogeneity.
For the prognostic meta-analysis, there was also heterogeneity (I 2 = 65:9%, p = 0:019). However, meta-regression analyses and subgroup analysis cannot be performed because of insufficient study numbers.
The present study has several limitations: firstly, the number of studies available for meta-analysis was small; secondly, the kind of ethnicity was monotonous; thirdly, the values of cut-off in the studies were partial and different; finally, it is difficult to conduct subgroup analysis, such as the influence of gender and tumor stage on the results due to the limited data of original articles.

Conclusions
In conclusion, this study demonstrates that miR-192 has a moderate diagnostic value to distinguish cancer patients from controls and also can be a promising positive prognostic biomarker in some types of cancer.

Data Availability
All data generated or analyzed during this study are included in this article. More information concerning the data can be obtained from the corresponding author.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this article.